Does Alteration of the Microbiome Cause Diverticular Disease?

Abstract

The role of the microbiome in influencing the development and course of human disease is increasingly understood and appreciated. In diverticular disease, the microbiome presents an intriguing potential link between the disease and its long-established risk factors, dietary fiber and industrialization. However, current data have yet to demonstrate a clear link between specific alterations in the microbiome and diverticular disease. The largest study of diverticulosis is negative and studies of diverticulitis are small and heterogeneous. Although multiple disease-specific hurdles exist, the early state of the current research and the many un- or underexplored clinical phenotypes present a significant opportunity for investigators to improve our knowledge of this common and incompletely understood disease.

The role of the colonic microbiome in the pathogenesis of diverticular disease is controversial. For the surgeon, source control of pathogenic bacteria released into the mesentery, peritoneal cavity, or bloodstream by diverticular colonic perforation is critical and achieved by antibiotics, percutaneous drainage, colectomy, and/or intestinal diversion as required by the patient's disease pattern and severity of illness. However, whether the innate intestinal microbiota contributes to the development of diverticula, the progression to symptomatic diverticular disease, or the severity of diverticulitis is unclear. This knowledge gap has therapeutic implications. Antibiotics, long the anchor of uncomplicated diverticulitis treatment, fail to improve short- or long-term outcomes compared with placebo in multiple randomized controlled trials, ^{1 2} suggesting controlling intestinal pathogens is not therapeutic in early disease and that antibiotics have been long overused. However, the potential for a role of the microbiome in diverticulitis is intriguing. Long-standing associations between fiber intake, industrialization, and diverticular disease could be explained by the microbiome, which is also strongly influenced by these factors. Compelling data in other gastrointestinal diseases and emerging basic research suggest a potential role for the microbiome in diverticulitis.

Challenges for microbiome research in diverticular disease include a variety of presentations, from a ubiquitous and occult precursor lesion to inflammation, hemorrhage, perforation, stenosis, and somewhat controversial chronic uncomplicated symptomatic disease states. Additionally, intermittent antibiotic treatments, confounding lifestyle factors, different microbiome analysis methodologies, lack of an animal model, and the relatively understudied nature of the disease make microbiome research in diverticulitis at its infancy compared with colon cancer and inflammatory bowel disease (IBD).

The Role of Fiber

Low fiber intake due to Western diet, leading to constipation, and ultimately structural changes in the colon is the classic and longest-standing risk factor for diverticular disease. ³ Fiber's complex and critical link with the microbiome makes the intriguing potential case for the microbiome being the link between fiber and diverticular disease. Along with resistant starch and oligosaccharides, dietary fiber makes up the majority of the nutrient sources for the colonic microbiota. Fiber is a collective term for a wide variety of complex, three-dimensional carbohydrate polymers that demonstrate high levels of diversity in chemical bonds, attachment to proteins, three-dimensional structure, and degree of fermentability. ⁴ A fiber's enzymatic digestibility is predicated by its discrete structure and is bacterial strain-dependent. For instance, a select few of Bacteroides spp. can digest complex arabinoxylans. Conversely, no species of Prevotella spp. can do so. Differences like these create competitive niches for individual species of bacteria and raise the possibility that specific components of diet can influence or be modulated to influence the composition of colonic microbiome. ^{5 6}

Lack of fiber intake is clinically associated with multiple common diseases, which may be due to altered microbiota. When investigating synthetic microbiota in gnotobiotic mice with 14 species of bacteria, the absence of fermentable fiber resulted in the preferential production and use of colonic mucin as a bacterial food source. After 2 weeks of a diet without fermentable fiber, there was a notable decline in Bacteroides ovatus and Eubacterium rectale . Conversely, there was a significant increase in the amount of Akkermansia muciniphila and Bacteroides caccae , two microbes responsible for mucin degradation and subsequent colonic wall thinning. ⁷ Despite their limitations, data like these represent a link between fiber, alterations to the colonic microflora, and structural change to the colon wall.

Development of Diverticulosis

Diverticula are noted on screening endoscopies in roughly 33% of asymptomatic patients aged 50 to 59 and in 71% of asymptomatic patients over age 80. ⁸ Development of diverticulosis is associated epidemiologically with age, lack of physical activity, red meat, intake, low fiber diet, and smoking. ^{9 10} Like most common diseases, development of diverticular disease is influenced by both environmental and inherent factors. Twin studies estimate 40 to 53% of individual risk is due to heritable factors. ¹¹ Diverticulosis exhibits strong geographic/anatomic trends, with a strong predilection for the sigmoid colon in North America and Europe and a right-sided dominance in Asia. ⁷ Diverticulitis is commonly considered a disease of Western civilization due to increasing incidence in developed countries over time; as Asian populations adopt a Western diet, incidence of diverticulitis increases without an anatomic shift to the sigmoid. Notably, the vast majority of epidemiologic and microbiome studies have been performed in sigmoid diverticulitis in European-ancestry populations.

In terms of the role of the microbiome, current data do not support strong differences between the microbiota of individuals with and without diverticulosis. Jones et al analyzed the microbiome of 535 asymptomatic patients undergoing first-time screening colonoscopy. A total of 226 patients were found to have diverticulosis and 309 had no diverticular disease. The authors performed mucosal biopsies in the mid-sigmoid for patients without diverticulosis or adjacent to sigmoid diverticula. These were subject to 16S ribosomal ribonucleic acid (rRNA) sequencing. The authors first assessed the Shannon Diversity Index, which measures the amount of diversity in a community, and found a significant difference, but an effect size of < 1%. Analysis at each phylogenetic level revealed weak to no associations with case–control status. Ultimately, aside from decreased representation of Proteobacteria and Comamonadaceae in patients with diverticular disease, both at r -squared values of 2%, there was no evidence of any association between components mucosal microbiome and presence of diverticulosis. ¹² The authors concluded that strong differences in the microbiome between patients with and without diverticulosis were not present.

As will be seen in the remainder of the review, in contrast to much of the other published research, this negative study has a large sample size, straightforward objective phenotypes, consistent absence of antibiotic therapy, unbiased approach to assessment of microbiota (as opposed to analysis of selected species), and uses mucosal, rather than fecal microbiota, which are more likely to be in close contact with the colonic epithelium.

Similar to the above study, a prospective analysis of routine screening colonoscopy patients identified 19 patients with diverticula and 24 without diverticular disease. In this study, biopsies were taken from both sigmoid and transverse colon locations to permit both intraindividual and interindividual analysis. Firmicutes to Bacteroides ratio was used to define dysbiosis, and this metric was the same between groups with and without diverticula. Cosine distance analysis was used to compare microbial composition. There was low correlation between individuals, and no difference between case–control groups. Microbiome diversity between the sigmoid and transverse colon was similar between study groups. The group also assessed mucosal neutrophil and lymphocyte counts and these were similar between the two groups. Overall, it was suggested that gut microbiome did not play a major pathologic role in development of diverticula. ¹³

However, other studies have demonstrated microbial differences between patients with diverticular disease and healthy controls. Of 51 patients undergoing planned colonoscopy, 16 were found with diverticula and 35 were without any diverticular disease. Diverticular disease was defined as the presence of diverticula with or without symptoms. Reported symptoms were similar to irritable bowel syndrome (IBS) including abdominal pain, bloating, and change in bowel habits. Radiologic confirmation of diverticulitis was not required. Biopsies were performed in all patients looking specifically at burden of Enterobacteriaceae with a secondary goal of evaluating the relationship between demographics, socioeconomic status, lifestyle habits, and gastrointestinal symptoms, and the microbiome. In the patients with diverticula, biopsies were taken from intact, inter-diverticular mucosa in the descending colon. Microbial analyses were achieved with deoxyribonucleic acid (DNA) extraction and quantitative polymerase chain reaction (PCR) to estimate the amount of Enterobacteriaceae . Microbial diversity was estimated by calculation of richness and the Shannon–Wiener and Simpson's diversity indices. Ultimately, patients with diverticula were found to have a higher amount of Enterobacteriaceae present in their biopsies when compared with patients without diverticular disease. None of the studied lifestyle or socioeconomic qualities affected the amount of gut Enterobacteriaceae . ⁵

Transition from Diverticulosis to Diverticulitis

Roughly 10 to 25% of patients with diverticulosis can be expected to develop acute inflammation of the diverticula resulting in diverticulitis. The triggers that lead to diverticulitis remain incompletely understood. Bacterial dysbiosis has been posited as a risk factor; however, evidence is limited. Challenges include sample size (combined together the following studies include only 49 patients), phenotyping cases and controls, and variation in timing and method of microbiome sampling and downstream analysis. Unsurprisingly, given these challenges there is no consistent finding across studies regarding diverticulitis and the microbiome. Clinically, the randomized controlled trials demonstrating no benefit of antibiotics in improving short- or long-term outcomes, suggest that altering the colonic bacteria may have limited therapeutic benefit in treating uncomplicated diverticulitis, but whether microbial triggers initiate diverticulitis in the first place remains unknown.

Two studies have compared surgical specimens from diverticulitis patients undergoing elective resection. One study used adjacent undiseased tissue as a control and a second compared diverticulitis specimens to patients with other pathologies. These studies would ideally identify microbial alterations present in patients with recurrent disease. Finally, a third study used rectal swabs in patients with an acute episode of diverticulitis.

First, a retrospective cohort study analyzed nine patients with chronic, recurrent diverticulitis who underwent surgical resection of the disease colon. Following resection, their microbial burden was assessed in areas affected by diverticulitis and compared with adjacent nonaffected tissue. Microbial and fungal burden was determined using 16S rRNA and internal transcribed spacer gene sequencing. Subsequently, inferred metagenomics analyses were conducted to identify predictive metagenomes associated with diverticulitis and adjacent nonaffected tissue. While diseased tissue demonstrated thickened bowel wall and increased neutrophils within the lamina propria, adjacent tissue was normal appearing and with negligible neutrophilic infiltration. Ultimately, diseased tissue was found to have an overrepresentation of Microbacteriaceae and Ascomycota when compared with adjacent, unaffected tissue. Adjacent, unaffected tissue was associated with a preponderance of Pseudomonas and Basidiomycota . ¹⁴

In the second study, surgical colon resections from 34 patients, 21 with colorectal cancer, 9 with diverticulitis, and 4 with IBD, were compared for mucosal microbiota composition. There was no healthy control. The mucosal specimens underwent extraction of DNA for Gram-positive bacteria. Following extraction, PCR primers specific to bifidobacterial microbiota were used for amplification. Total bifidobacterial levels were calculated and the most frequently identified species were assessed. All nine of the diverticulitis patients and all four of the IBD patients had Bifidobacteria growth from their mucosa. Overall, diverticulitis patients had a higher total count of Bifidobacterium than colorectal cancer and IBD. Specifically, there was a significantly higher occurrence of B. longum in diverticulitis than patients with colon cancer or IBD. ¹⁵

Finally, as an alternative approach, rectal swabs from 31 computed tomography-confirmed left-sided uncomplicated Hinchey 1A/1B diverticulitis were compared with 25 patients without diverticular disease to assess for differences in the fecal microbiome. Control patients' rectal swabs were obtained before screening colonoscopy. DNA extraction was performed and subsequently amplified with two multiplex PCRs. The first PCR was for phyla Firmicutes , Bacteroidetes , Actinobacteria , Fusobacteria , and Verrucomicrobia . The second PCR amplified for the phylum Proteobacteria . Diversity analyses were calculated as the Shannon index. A partial least squares discriminant analysis (PLS-DA) regression model was used as a prediction of whether the sample was from a control or diverticulitis patient. PLS-DA provided an estimate of the discriminatory power of each descriptor. There was no difference in the Firmicutes to Bacteroidetes ratio in either group, pointing away from dysbiosis. The total load of Proteobacteria was similar between both groups. However, the Shannon index demonstrated that there was significantly greater diversity in the Proteobacteria phylum among diverticulitis patients when compared with healthy controls. Finally, per PLS-DA, the most discriminative species to predict past diverticulitis was from the family Enterobacteriaceae with a diagnostic accuracy rate of 84%. ¹⁶

Chronic Symptomatic Uncomplicated Diverticular Disease States

Compared with diverticulitis, there is relatively more published research on symptomatic uncomplicated diverticular disease (SUDD). SUDD is defined as symptoms including pain, bloating, and bowel habit changes in the presence of diverticula but in the absence of macroscopic inflammation. SUDD is a somewhat controversial clinical entity given the ubiquity of diverticulosis in older age groups and the diagnostic overlap with IBS, constipation, and diverticulitis itself. ¹⁷ Triggers of SUDD have been postulated including microscopic inflammation, visceral hypersensitivity, and altered motility. ¹⁸ Dysbiosis as a potential causative factor has also been queried in several studies with a total of 114 patients, described in the following paragraphs. In these studies, some authors supplemented their analysis with metabolomics and others attempted to correlate individual symptoms, such as pain or bloating, with microbiome analysis.

Thirty-eight patients who underwent screening colonoscopy were divided into a control group without disease (14), diverticulosis (16), and SUDD (8) and their colonic microbiota was analyzed. In this study, SUDD was defined as the presence of uncomplicated diverticula with abdominal pain and/or bloating with or without changes of bowel habits. In patients with diverticulosis and SUDD, four biopsies were obtained in the left colon, one biopsy was obtained within 1 cm of a sigmoid diverticulum, and two biopsies were taken 20 cm above the most proximal diverticulum. In control patients, four biopsies were obtained in the descending colon. Fecal and urine samples were also obtained. For fecal samples and mucosal biopsies, total bacterial DNA was extracted and amplified. Fecal and urine samples were analyzed with nuclear magnetic resonance spectroscopy (NMR). Ultimately, regardless of symptoms, the presence of diverticula was associated with an increased quantity of colonic macrophages on mucosal biopsies. There were no differences between the mucosal microbiota of healthy patients and those with diverticulosis in the immediate diverticular region or at distant sites. Patients with SUDD had a lower amount of Akkermansia in the immediate diverticular region as compared with a distant site. Additionally, patients with symptomatic disease were found to have a depleted quantity of Clostridium cluster IV, Clostridium cluster IX, Fusobacterium , and Lactobacillaceae in fecal samples. There was a negative correlation between quantity of macrophages and mucosal Clostridium cluster IV and Akkermansia . Six urinary metabolites were found to correctly and reliably exclude one of three diagnoses 95% of the time. ¹⁹

Other studies have not demonstrated such robust differences between healthy controls and patients with diverticular disease. For example, stool samples were obtained from a group of 44 women, 15 patients with SUDD, 13 with asymptomatic diverticulosis, and 16 healthy patients. SUDD was defined as known diverticulosis with any history of left lower quadrant pain for over 24 hours but without any complications such as stenosis, abscess, or fistula. Stool studies were obtained at least 4 weeks following colonoscopy. Bacterial DNA was extracted and amplified with PCR to quantify total bacteria. Primers were utilized for Bifidobacterium and Lactobacillus genera as well as Clostridium coccoides , Bacteroides-Prevotella , Escherichia coli , and Akkermansia muciniphila . Additionally, fecal metabolome was assessed with NMR. With analysis, PLS-DA was utilized for metabolic profiles. Ultimately, there was no difference in overall bacterial quantity between the three groups. However, in contrast to the above study healthy patients were found to have less Akkermansia muciniphila than patients with asymptomatic and symptomatic diverticulosis. SUDD patients demonstrated greater amounts of N-acetyl-glucosamine and U1 than asymptomatic patients. ²⁰

Another study aimed to compare the fecal microbiome of patients with SUDD, defined as diverticulosis with abdominal pain and/or bloating, to that of patients with past diverticulitis. From stool samples, calprotectin was quantified and DNA was extracted. Alpha diversity measurement was accomplished by taxon richness, Shannon diversity index, Simpson evenness index, and Good's coverage. Of the included patients, 15 had no prior history of diverticulitis; whereas, an additional 15 did have a prior episode of diverticulitis. Comparing those two groups, there was no difference in α diversity or microbiota composition. However, some taxonomic groups did demonstrate differences in observed relative abundance. Specifically, those with previous acute uncomplicated diverticulitis were found to have differences in the abundance of Pseudobutyrivibrio and Bifidobacterium . Christensenellaceae family, and Mollicute RF9 order were more prevalent in patients without a history of diverticulitis. A higher quantity of Ruminococcus was found to be positively correlated with bloating score; whereas, bloat score was inversely related to abundance of Roseburia . Patients with an increased abundance of Cyanobacterium were identified to have higher pain scores. ²¹

Finally, a separate study compared four patients with diverticular disease to eight controls to assess for differences in fecal microbiota. Diverticular disease patients were termed uncomplicated but there was no specification regarding type or frequency of symptoms. Healthy control patients had no known intestinal pathology and were of a normal body mass index. Stool samples were obtained from all patients and microbial DNA was isolated. Preselected primers were used to amplify the hypervariable regions V1-V3 of the 16s rRNA gene and the amplification products were subsequently purified. Operational taxonomic units (OTUs) were selected and taxonomy was assigned to compared OTUs to the Greengenes database. Alpha diversity measures were calculated. There was a depletion in Bacteroides fragilis in patients with diverticular disease; however, there was no overgrowth of any one lineage nor increased overall quantity. ²²

Potential Next Steps

To further clarify the role of the microbiome in diverticulitis, future studies should attempt to overcome the challenges observed in this review: heterogeneity of study design, small sample sizes, variable diverticular disease phenotyping, and collection of samples at disease-relevant intervals. Additionally, many diverticular disease states remain unstudied: right-sided diverticulitis, diverticular hemorrhage, small bowel diverticulitis, stenosis, and fistula. The alterations to the microbiome, or lack thereof, caused by sigmoidectomy remain unknown as well. These knowledge gaps leave significant potential opportunities for future investigators.

Major hurdles remain. Patients with no diverticula included as controls may just be in a “pre-diverticular” state, biasing diverticulosis studies to the null. Lack of risk prediction tools makes it impractical and costly to prospectively follow the microbiome of asymptomatic diverticulosis patients to study which few will develop diverticulitis over time. During acute diverticulitis, mucosal sampling risks perforation, limiting sampling methodology. Finally, lack of an animal model of diverticular disease makes critical follow-up basic research impossible. To surmount these challenges, robust collaboration between surgeons, endoscopists, patients, and scientists is required to inform basic research with clinical relevance.

Conclusion

The role of the microbiome in diverticular disease is currently controversial. The potential of the microbiome to explain the long-standing associations of diverticular disease, diet, fiber, and industrialization is compelling. However, the largest study of diverticulosis to date is negative. Studies of the microbiome in diverticulitis are small and lack clear case–control differences, and a preponderance of current research is focused on somewhat controversial chronic symptomatic uncomplicated disease states. Evidence-based clinical practice is moving away from routine antibiotics. While the current data lack a clear causative role for the microbiome, the potential for further research is incredible with abundant opportunity for clinical investigators to improve the science and challenge current understanding of this common, but understudied disease.